Drift characteristics of Sea-Bird dissolved oxygen optode sensors

Abstract The Sea-Bird 63 dissolved oxygen optode sensors used on various oceanographic platforms are known to drift over time. Corrections for drift are necessary for accurate dissolved oxygen measurements on the timescale of months to years. Here, drift on 14 Sea-Bird 63 dissolved oxygen optode sensors deployed on Spray underwater gliders over 5 years is described. The gliders with oxygen sensors were deployed regularly for 100-day missions as part of the California Underwater Glider Network (CUGN). A laboratory two-point calibration was performed on the oxygen sensor before and after glider deployment. Sensor drift during 100-day deployments was larger than during 100-day storage periods. Sensor behavior is modeled with a gain that asymptotes to 1.090 ± 0.005 with an e-folding timescale of 3.70 ± 0.361 years. At zero oxygen concentration, the sensor consistently reads around 3 μmol kg −1 ; a negative offset term is used in addition to the gain to correct the sensor oxygen. The correction procedure removes the error due to long time drift, one of the major sources of error, with an uncertainty of 0.5% (0.9% including outliers) or 0.5 μmol kg −1 depending on concentration, which improves the accuracy of the Sea-Bird 63 although uncertainty from other sources of error including the initial factory calibration and the sensor response time remain. Suggested procedures for implementing a two-point calibration procedure in the laboratory are discussed. Calibrations must be considered starting 6 months after initial factory calibration to keep error from sensor time drift under 1%.